Using methods from the data-mining and machine-learning literature for disease classification and prediction: a case study examining classification of heart failure subtypes.

摘要: Abstract Objective Physicians classify patients into those with or without a specific disease. Furthermore, there is often interest in classifying according to disease etiology subtype. Classification trees are frequently used the presence absence of However, classification can suffer from limited accuracy. In data-mining and machine-learning literature, alternate schemes have been developed. These include bootstrap aggregation (bagging), boosting, random forests, support vector machines. Study Design Setting We compared performance these methods that conventional heart failure (HF) following subtypes: HF preserved ejection fraction (HFPEF) reduced fraction. also ability predict probability HFPEF logistic regression. Results found modern, flexible tree-based literature offer substantial improvement prediction subtype regression trees. had superior for predicting proposed literature. Conclusion The use offers over subtypes population-based sample Ontario, Canada. do not improvements HFPEF.